It is widely recognized that antibiotics are overprescribed in children. Nearly 67 million pediatric antibiotic prescriptions were written in 2013 in the United States, with children in the South being more likely to receive an antibiotic compared to those in the West (952 vs. 555 prescriptions per 1,000 children).1
Up to half of acute respiratory infections are unnecessarily treated with antibiotics.2 Among children, acute otitis media (AOM) is the most common condition for which antibiotics are prescribed.2 Antibiotic treatment for AOM in children produces a benefit limited,3–5carries a risk of adverse events6 and contributes to the growth of antibiotic resistance.7–9
There has been growing support for watchful waiting approaches in the management of AOM. Watchful waiting occurs when a doctor chooses to observe a patient for 2 to 3 days after AOM diagnosis to determine the need for antibiotics.
The 2004 American Academy of Family Physicians and American Academy of Pediatrics (AAP) practice guidelines were one of the first to promote watchful waiting as a treatment option for AOM in children.10 Likewise, many European AMO guidelines emphasize a watchful waiting approach.11 Watchful waiting has been shown to decrease antibiotic prescribing rates and caregivers appear to be satisfied with this approach if it is adequately described.5,12,13
Multiple retrospective cohort studies have evaluated the impact of the 2004 guidelines and found no change in antibiotic prescribing in the United States, with up to 85% of cases receiving antibiotics.14–17 Similarly, European studies have found that More than 80% of AOM episodes are initially treated with antibiotics.18–20 Suggested reasons for such high rates of antibiotic use include parental pressure,21,22 differences in training and clinical experience,23– 25 patient race,26 age, misdiagnosis of AOM,27 and physicians’ desire to prevent complications of untreated AOM.28
The most recent AAP AOM treatment guideline published in 2013 was more explicit than the 2004 guidelines about who should receive antibiotics immediately.29 Specifically, the 2013 guideline defined AOM and excluded previous signs and symptoms criteria. symptoms that may have resulted in misclassification of upper respiratory tract infections as AOM.30
The recommendations emphasized 2 landmark studies that used strict diagnostic criteria for AOM in younger children and compared treatment with amoxicillin-clavulanate versus placebo.31,32 Hoberman 2011 found no difference in initial resolution of symptoms between the 2 groups, while Tahtinen 2011 found a significant decrease in treatment failure (defined as no improvement in performance, worsening of symptoms or other complications) in the antibiotic group compared to placebo (hazard ratio (RR) 0.38, range 95% confidence interval [CI] 0.25–0.59). Both studies also found an increased risk of adverse reactions for the treatment group compared to the placebo group.
It is unknown whether the new AOM guidelines resulted in a change in antibiotic prescribing. This study aimed to determine treatment trends for uncomplicated, nonrecurrent AOM in a commercially insured pediatric population in the United States. Additionally, determinants of watchful waiting, including physician prescribing tendency, were evaluated to elucidate what drives AOM treatment decisions.
| Methods |
> Data source and study population
Patient cohorts were established from enrollment and billing records in the IBM MarketScan commercial claims research databases (2005 to 2019). MarketScan provides longitudinal information on healthcare utilization for a national sample of privately insured employees, retirees, and their dependents in the United States. Data include diagnoses and procedures associated with outpatient and inpatient visits, as well as outpatient pharmacy medication dispensing claims. This study was reviewed by the University of Florida Institutional Review Board and approved as exempt due to the use of de-identified data.
AOM episodes were identified based on the primary diagnosis from outpatient visits according to the International Classification of Diseases, Ninth Revision, Clinical Modification (381.0x, 382.0x) or ICD-10-CM (H65.19x, H66.00x ). AOM episodes in children aged between 1 and 12 years at the time of diagnosis were included. Continuous enrollment in comprehensive health insurance and prescription drug plans was required for 12 months before and 1 month after each episode of AOM.
To capture episodes of uncomplicated nonrecurrent AOM, episodes that were preceded by another visit with a diagnosis of AOM within the previous 6 months were excluded. Accordingly, patients could contribute multiple episodes of AOM if the episodes were spaced more than 6 months apart. Also excluded were AOM episodes preceded by diagnoses of chronic otitis media with effusion, myringotomy or perforation of the tympanic membrane in a 6-month retrospective period or that were accompanied by a diagnosis of otitis externa in the same consultation as the AOM episode or which resulted in the prescription of a non-oral antibiotic within 7 days.
Episodes following tympanostomy tube placement identified through procedure codes in the past 12 months were also excluded.33,34 Finally, episodes with an inpatient or outpatient medical visit that included an acute infection were excluded. as a primary or secondary diagnosis within 2 weeks before and 1 week after the index AOM diagnosis to help ensure that antibiotic use was for AOM and not another infection.
> Study result
Using pharmacy claims, all oral prescriptions for penicillins, cephalosporins, macrolides, tetracyclines, quinolones, sulfonamides, and other β-lactams up to 7 days after AOM diagnosis were extracted. Episodes were grouped into 2 groups based on treatment status. Those with a prescription within 3 days of AOM diagnosis were assigned to the treatment group, while those without a prescription or with prescriptions filled within 4 to 7 days of diagnosis were assigned to the watchful waiting group. Finally, any episode was excluded from analysis upon a negative pharmacy claim regarding days supply, quantity, or copayment to ensure accurate classification of each AOM episode as managed with antibiotics or watchful waiting.
> Covariates
The authors examined the effect of both patient level and physician characteristics on AOM treatment decisions. For patients, age at the time of AOM diagnosis, sex, geographic region, diagnosis of concurrent fever in the index episode, conditions associated with more severe AOM episodes,35–37 type of health insurance plan, month and year of diagnosis, and ranking of care quality of the state of residence. For this, a health quality variable extracted from the US News Portal was used to assign patients to the first 10 positions, the last 10 or the middle positions of the medical care quality ranking.38
Information on the physician’s specialty was verified from the consultation that defined the index episode of AOM. Physician prescribing tendency was defined using 2 different methods.
The study cohort was first restricted to AOM episodes that were diagnosed by physicians with at least 30 AOM episodes in the database. Physicians who treated at least 80% of all their AOM episodes (excluding the index AOM episode) were considered high-volume antibiotic prescribers, while physicians whose AOM episodes resulted in 20% antibiotic prescribing or less within 3 days of diagnosis were classified as low-volume prescribers of antibiotics.39
Second, to extend the analysis to physicians with less frequent AOM visits in the database, prior treatment decisions were captured for each physician within the cohort identified based on date of diagnosis to characterize the trend toward prescription; For example, a doctor who treated the previous episode was considered to have a tendency to treat.40
> Statistical analysis
The proportion of patients with an episode of AOM who had a completed antibiotic prescription within 3 days was plotted during the study months, where an episode was assigned to a month based on its date of diagnostic consultation. The prevalence of AOM episodes among pediatric patients aged 1 to 12 years was also reported in each study month to provide context on the overall impact of AOM treatment. Finally, multivariable logistic regression models were used to examine all measured determinants of watchful waiting.
> Sensitivity analysis
To evaluate the impact of bilateral infections, a sensitivity analysis was performed using AOM episodes diagnosed in the ICD-10 era, which has more granular ear-specific diagnostic codes. The multivariable logistic regression model was rerun to examine the determinants of watchful waiting, adding this new covariate. Data analyzes were performed using SAS version 9.4 (Cary, North Carolina). The secular trend graph was created with R version 4.0.3.
| Results |
2,176,617 primary AOM episodes were identified, of which 1,693,690 (77.8%) continued to fill a prescription within 3 days. Among the clinical criteria for exclusion of AOM episodes, a history of AOM in a 6-month retrospective period was the most common, resulting in the exclusion of one-third of all episodes (1,618,125, 34%).
The average age of children contributing to an episode was 4 years, and the largest proportion resided in the southern region of the United States and were enrolled in a restrictive health plan. About two-thirds of the episodes were diagnosed by a pediatrician. Compared with physicians from other specialties, otolaryngologists were the only ones with clearly different representation between treatment groups with 72.2% of episodes assigned to the watchful waiting group.
Throughout the study period, 51,637 physicians who treated at least 1 episode of AOM and 20,259 physicians who treated at least 2 were identified. Among episodes managed by physicians who treated the previous episode within the cohort of AOM episodes in this analysis, 81.6% were treated, while only 66.8% received treatment when seen by physicians who had not treated the previous episode.
Restricting AOM episodes to those diagnosed by physicians who had managed at least 30 episodes (n = 4377) during the study period resulted in about a quarter (625,576) of all eligible AOM episodes available for analysis. A total of 2261 (51.7%) of these physicians were classified as high-volume prescribers of antibiotics (i.e., treating ≥80% of episodes), and 90 (2.1%) were categorized as low-volume prescribers. volume of antibiotics (treatment of <20% of episodes).
Among episodes assigned to the first group of prescribers, 318,441 (85.3%) were treated, while this proportion was reduced for episodes managed by physicians with low volume of antibiotic prescriptions (71.9% and 12.1% for episodes managed by physicians who treated between 21% to 79% or 20% or fewer episodes, respectively).
Treatment prevalence remained constant throughout the study period with 77.8% and 84.1% of AOM episodes treated in the first and last month of the study, respectively. Treatment patterns showed some seasonality with a reduction
of the prevalence of antibiotic use during the warmer months. These results were consistent across medical specialties with otolaryngologists showing lower antibiotic use throughout the study period. The prevalence of AOM episodes showed strong seasonality, but remained similar throughout the study period, with 49 and 60 episodes per 10,000 pediatric patients enrolled in February 2006 and 2019, respectively.
Amoxicillin, cephalosporins and amoxicillin/clavulanate were the most common antibiotics used for both the treatment group and the small proportion of episodes (2.8%) managed with watchful waiting that resulted in a prescription indication between 4 and 7 days of AOM diagnosis. Amoxicillin was used more commonly in the treated group (52.6% vs. 31.3% in the treated watchful waiting group).
In the treatment group, 1,638,651 (96.8%) episodes had their antibiotic prescription on the same day as the AOM diagnosis. To examine whether caregivers could have filled prescriptions immediately, even if they followed expectant management recommendations, we assumed that caregivers would be reluctant to pay for a medication when it is eventually not necessary. It was found that most prescriptions required a copayment and that the median copayment for the treated watchful waiting group was similar to that of the treatment group.
In the primary multivariable analysis restricted to AOM episodes that were managed by physicians with at least 30 episodes in the data sets, a diagnosis of failure to thrive (odds ratio [OR] 1.47, 95% CI 1.47) was found to be 24–1.74) and atopic dermatitis (OR 1.09, 95% CI 1.02–1.16) increased the odds of watchful waiting for AOM.
AOM episodes with a diagnosis of concurrent fever were less likely to be managed with watchful waiting than those without (OR 0.63, 95% CI 0.59–0.68), as were those with a diagnosis of cancer. Patients seen by otorhinolaryngologists were more inclined to watchful waiting compared to those seen by a pediatrician (OR 5.45, 95% CI
5.21–5.710). Other physicians were less likely to use watchful waiting compared with pediatricians (internal medicine, OR 0.89 [95% CI 0.83–0.96]; emergency medicine, OR0.80 [95% CI 0.74 –0.87]; or family medicine, OR 0.84 [95% CI 0.81–0.86]). Episodes during the summer months more likely resulted in watchful waiting. Looking at annual trends, a lower propensity for watchful waiting was found each year between 2014 and 2019 compared to 2006 (e.g., for the comparison from 2019 to 2006 OR 0.76, 95% CI 0.73-0.79 ).
Finally, physicians who treated less than 20% of their episodes (low-volume prescribers of antibiotics) were highly likely to use watchful waiting for the assessed AOM episode compared to physicians who treated 80% or more (OR 11 .61, 95% CI 10.66-12.64).
Defining prescribing trend based only on the treatment decision of the previous AOM episode within the cohort yielded similar results, including the seasonal nature of treatment, as well as the maintenance of prescribing trend as the strongest predictor for watchful waiting. when comparing doctors who did not treat the previous episode with those who did (OR 1.83, 95% CI 1.80-1.85).
Using all available AOM episodes for analysis, omitting physician prescribing tendency as a covariate, increased the effect of medical specialty, indicating an association between specialty and prescribing tendency. Otorhinolaryngologists compared to pediatricians were more likely to use watchful waiting (OR 9.16, 95% CI 8.99-9.34).
Other physicians were less likely to adopt watchful waiting compared with pediatricians (internal medicine, OR 0.87 [95% CI 0.84–0.89]; emergency medicine, OR 0.79 [95% CI 0.89] 76–0.82]; or family medicine, OR 0.80 [95% CI 0.79-0.81]). Sensitivity analysis yielded similar results for prescribing tendency and clinical specialty (specifically otolaryngology) as the strongest predictors for watchful waiting. Compared with AOM episodes explicitly diagnosed as unilateral, bilateral infections were more frequently managed with watchful waiting (OR 1.29, 95% CI 1.25–1.33), as were AOM infections with nonspecific coding (OR 1.92, 95% CI 1.81-2.04).
| Discussion |
This study provides updated evidence on the treatment of AOM relevant to national efforts to reduce unnecessary antibiotic prescribing, including the potential impact of the AAP AOM guideline published in 2013. This guideline updated the diagnostic criteria for AOM. and provided explicit criteria for immediate antibiotic treatment of AOM in children.
For non-severe cases, it gave the option of watchful waiting as a treatment approach as long as follow-up care could be provided, offering analgesics for pain management, and with shared decision making with caregivers.30 These recommendations were based on several clinical trials with strict AOM diagnostic criteria that compared antibiotics with placebo and revealed a modest benefit of antibiotics for the resolution of symptoms and an increased risk of adverse effects.31,32 Subsequent studies and meta-analyses have also demonstrated a minimal benefit with immediate treatment of AOM.41,42
Three key findings from this study are notable: First, a small increase in treatment prevalence was found over the study period, with 84.1% of uncomplicated, nonrecurrent AOM episodes followed by prescription of antibiotics within 3 days in the last month of study.
Second, the strongest predictors of watchful waiting were the physician’s clinical specialty and prescribing tendency, even when the tendency was loosely defined as the most recent preceding treatment decision. In contrast, patient demographics and clinical characteristics played a minor role.
Third, only a small proportion of patients (2.8%) who appeared to be managed with a watchful waiting approach had an antibiotic prescription after the first 3 days of diagnosis.
This third finding strongly supports the self-limiting nature of most AOM episodes and is consistent with an Italian study that reported 5.6% of patients using a prescription after a period of watchful waiting.20 Other studies have similarly concluded that the majority of treated AOM episodes could have been managed with watchful waiting.4,5,43-46
When patients in the watchful waiting group finally completed an antibiotic prescription, amoxicillin was the least common first-line agent, unlike patients who received immediate antibiotic therapy. The authors suspect that second-line agents were often more favored in this select group as the episodes were not de facto self-limiting.
The authors’ findings regarding the persistent trend toward immediate antibiotic treatment are consistent with previous studies, which have shown no effect of previously published guidelines.19,20 Recent studies from Europe and Israel have reported that up to 80% of episodes of AOM are treated immediately with antibiotics, data comparable to the findings of the present work.18,19,47-49 Although certain clinical scenarios could have justified immediate antibiotic treatment, a wide range of risk factors for AOM were intentionally excluded. recurrence and complications of AOM. This should have biased the study population to those with less need for immediate antibiotic treatment, resulting in an underestimation of the tendency towards immediate antibiotic prescribing in current practice. This study, considered alone and in conjunction with reports from other countries, suggests that current guidelines need to be revised to effectively reduce unnecessary antibiotic use in AOM.
Patient factors dictate the considerations for AOM management in the AAP guideline and the clinical trials from which it is derived. However, patient factors appeared not to be the main drivers of antibiotic prescription for uncomplicated nonrecurrent AOM in the present study. For example, fever and bilateral AOM were highlighted in the AAP guideline as criteria to support antibiotic treatment. Although fever was associated with a higher likelihood of antibiotic treatment, bilaterality was associated with a higher likelihood of watchful waiting. Similarly, developmental delay, immunodeficiencies, and many comorbidities that could exacerbate the severity of AOM were associated with a higher likelihood of watchful waiting or had little influence on the antibiotic treatment decision of the AOM episode.
On the contrary, factors not directly related to the patient were closely linked to treatment tendency. For example, AOM in the summer months was more likely to be treated with watchful waiting, despite the lack of association between season and AOM severity. The impact of clinical factors, both specialty and clinical prescribing tendency, was notable because these factors were much stronger than patient factors.
The severity of a previous case of AOM should not predict the severity of AOM in a subsequent patient or the need for antibiotic treatment. Thus, clinical practice patterns tend to drive the management of non-severe AOM more than patient factors. In other words, the likelihood that a patient will receive antibiotics depends more on which doctor they see than on the patient’s clinical characteristics.
It is well known that practice patterns, particularly for upper respiratory tract infections, differ by specialty.2,50-52 Both otolaryngologists and pediatricians may have more experience evaluating children’s ears than other primary care professionals, which which could explain their more judicious use of antibiotics, a fact compatible with previous reports.23,24
The varying prescribing practices across clinical specialties may also indicate that the AAP guideline has not been sufficiently disseminated to all physicians who regularly treat episodes of AOM. Up to 40% of identified AOM episodes were managed by specialties other than pediatricians. This may require greater dissemination of best practices that have successfully reduced antibiotic use.53-56
Parental factors, concerns about complications, and the need for follow-up care have been reported as major barriers to the adoption of watchful waiting.28 New research should evaluate what factors determine the adoption of watchful waiting, especially among those physicians who seem handle most cases with this approach. Importantly, comparing AOM management outcomes between clinicians who tend to prefer versus reject watchful waiting in real-world clinical practice would provide critical evidence that, if supportive, may offer the necessary assurance that Watchful waiting carries an acceptable risk-benefit.
Several limitations are recognized in this study. First, the study cohort, drawn from a nationally representative sample of commercially insured patients in the United States, is not generalizable to publicly insured patients who may be seen by different physicians and have different AOM risk profiles.
Second, the attempt to focus on patients with uncomplicated AOM may have been too restrictive. For example, recurrent uncomplicated AOM (commonly defined as 3 episodes within 6 months) is common and is not an indication for treatment alone, but such cases were excluded in this study to focus on the greater opportunities to implement the watchful waiting. On the other hand, the authors may not have been able to capture all the constellations of risk factors (e.g., history of general illness and presentation of AOM or sociodemographic factors, including parental pressure28) that a clinician may have considered. to justify treatment.
Third, these results are based on the assumption that the diagnosing physician also wrote the prescription, which is considered likely since AOM episodes that overlapped with other acute infections were eliminated. Caregivers were also trusted to follow instructions to wait to use the prescription if they were included in a watchful waiting approach. One study found that 27% of parents report noncompliance with watchful waiting when a prescription is provided at the time of diagnosis57; therefore, the prevalence of immediate treatment reported by the authors may be an overestimate. However, it was noted that most patients paid for the prescription out of pocket, which may reduce willingness to obtain a medication that may not be necessary. Additionally, patients who received a price discount for their prescriptions without using insurance could not be considered, information that may not be captured in the claims data.
Fourth, the database did not allow the distinction between observation as a medical practice that defines watchful waiting and delay in prescribing or filling previous prescriptions, which may occur for other reasons. Finally, this study did not aim to evaluate the impact of the new OMA guidelines on diagnostic accuracy. Assessment of diagnostic accuracy requires more clinical detail available in claims data, and assessment of trends should consider overall secular changes in the epidemiology of infectious diseases and other factors, such as increased acceptance of pneumococcal vaccination.58 59
| Conclusions |
We found no impact of the 2013 AAP guideline on the management of non-recurrent uncomplicated AOM, which allows for the option of watchful waiting before using an antibiotic. The strongest predictor of watchful waiting was provider prescribing bias and medical specialty, with limited impact of patient characteristics on treatment decisions.
Given these findings, more research is needed to understand what motivates doctors to take the watchful waiting approach and how outcomes in their patients compare to those who defer antibiotic use.
| Comment |
A large percentage of acute respiratory infections are unnecessarily treated with antibiotics; In children, acute otitis media is one of the most common prescribing conditions.
The indiscriminate use of antibiotics in these situations may offer minimal benefit and increase the risk of adverse effects and bacterial resistance.
Recent evidence suggests a watchful waiting approach to the management of non-recurrent uncomplicated AOM, in order to make rational use of medications and decrease complications associated with indiscriminate use.
For this approach to be applied at a general level, it will be necessary to have updated AOM management guides and training of professionals who care for children in various settings, in order to provide clear and concise information to parents, follow-up care and prescription of medications. antibiotics when considered really necessary.
